摘要
为了提高商品虚假评论的识别效果,提出了一种基于习惯偏差与xgboost算法的虚假评论识别方法。首先,通过提出新的算法来计算情感极性,同时在位置因素的基础上加入本地化情感词,从而提高评论文本情感极性计算的精准度。然后,提出新的用户习惯偏差指标和商家异常波动区间值并将其与几维重要特征融合在一起,进而得到一个关于评论-评论者-商户三者特征的新模型。最后,再与xgboost算法进行结合完成虚假评论的检测。实验结果证明,其能更有效识别虚假的评论信息,为消费者提供更加安全有价值的参考信息。
In order to improve the effect of recognizing fake merchandise reviews,the paper proposed a method for fake merchandise reviews recognition based on habit deviation and xgboost algorithm.First of all,an improved algo-rithm for emotional polarity calculation was proposed,and the location factors were added to the localised emotion words to make the calculation more accurate for the emotional polarity of the comment text.Secondly,the abnormal fluctuation interval of the merchant and the customary deviation index of the comments were put forward and the fea-ture model between reviews,reviewers and merchants were obtained with the other features.Finally,we worked to detect fake reviews combining with the xgboost algorithm.Experiments show that this method can effectively identify the fake reviews and provide more effective reference information for consumers.
作者
周娅
吴昱翰
ZHOU Ya;WU Yu-han(School of Computer and Information Security,Guilin University of Electronic Technology,Guilin Guangxi 541004,China)
出处
《计算机仿真》
北大核心
2020年第1期473-477,共5页
Computer Simulation
基金
国家自然科学基金委(61662015)
广西科技厅科技开发重点项目(桂科攻1598019)
NSFC-广东联合基金重点项目(U1501252)。
关键词
用户习惯偏差
店铺异常波动
情感极性
虚假评论
User's habit deviation
Abnormal fluctuation of shop
Emotional polarity
Fake Review
